The front of the Testudo

Van der Poel started building this machine in 1947 during his time at the Delft Technische Hogeschool. It was made for the optical department to deal with special optical calculating problems. There was no budget to work with, but luckily Van der Poel had by now become acquainted with Kosten and so he obtained 600 discarded relays, for free. They all had to be rewound and Van der Poel did that himself using a lathe.

Relay trouble

In the initial experiments with the relays Van der Poel ran into the same problems as Scholten en Loopstra, he initially regarded 30 milliseconds ample time for the relays switching time, but in practice, due to the occurrence of bouncing it often became around 900 milliseconds.


As Van der Poel graduated and left to work at the PTT with Kosten other students had to continue in the makeshift low budget environment. The machine was never really finished, though it was used more or less regularly from some time between 1952 and 1954 until 1965. It wasn't a fast machine: it did the calculations about as fast as a human could, but it was reliable and could be set to work overnight without supervision.

As it was a machine to work on optical calculating problems it would often have to solve sqrt(1-x^2). That calculation was eventually built into the machine, simply because it was used so often[1].


  • Developed at the Delft Technische Hogeschool
  • Building started in 1947 by Van der Poel
  • Never really completed
  • Programmable, digital, relay computer
  • Regularly used from sometime between 1952 and 1954 till around 1965
  • Designed for optical calculating problems for the optical department
  • TESTUDO is latin for turtle
  • Didn't calculate much faster then a human
  • Very reliable
  • Had sqrt(1-x^2) built-in as it was a very common operation


  1. Kranakis, E. , "Early Computing in the Netherlands", CWI Quarterly, vol. 1, issue 4, pp. 61-84, 12/1988.
Previous Image 1/2 NextTestudo (Front)Testudo (Front)The front of the Testudo.